Genomic based studies of disease now involve highly diverse types of data collected on large groups of patients. A major challenge facing scientists is how best to combine the data, extract important features, and comprehensively characterize the ways in which they affect an individual's disease course and/or likelihood of response to treatment. This project aims to develop statistical methods to address important problems that arise in genomic based studies of disease. In particular, we propose methods to improve the power and accuracy of results obtained from genome-wide studies of gene expression. We also propose statistical methods that integrate data across multiple platforms and scales. These integrative methods enable powerful inference related to identifying and quantifying groups of features that change across biological conditions (e.g. healthy vs. disease), and they also allow for the identification of important collections of features that affect a patient's disease course and/or treatment response. Successful completion of the project will help to ensure that maximal utility is gained from the powerful genomic-based technologies that are now routinely used in efforts to gain insights into and information about the genomic mechanisms underlying disease manifestation, progression, and maintenance. PUBLIC HEALTH RELEVANCE: The development of statistically sound approaches to resolve the genomic basis of complex traits is vital to individualizing medicine and improving public health. Ideally, high-throughput genetic, genomic, and pheno- typic measurements on diseased individuals would lead quickly to the identification of the salient features underlying their disease, along with a specification about how these features affect disease course. Many challenges in biostatistics must be overcome before this ideal is achieved. This proposal addresses some of those critical challenges.